Scholarlink

<p>In the practical application of rolling bearing fault information extraction technology, entropy is considered to be a promising feature extraction tool. The traditional entropy methods are vulnerable to environmental noise, which complicates the precise extraction information and results in low fault recognition rates. For this reason, the concept of symbolization is introduced based on entropy theory, using a differential evolution algorithm and an incremental algorithm to enhance method’s performance. the proposed method is called normal cumulative increment arc length entropy (NCIAE). To further obtain more feature information, we suggest extending NCIAE to a multiscale analysis, termed multiscale normal cumulative increment arc length entropy (MNCIAE). Finally, different performance testing models and two operational fault bearing datasets have been used to validate the proposed method. Experimental results demonstrate that this method not only higher computational efficiency but also outperforms traditional methods such as good consistency and robustness, achieving the highest classification accuracy.</p>

错误：搜索内容不能为空，请输入英文关键词
错误：关键词超出字数限制，请精简
高级检索

Normal cumulative increment arc length entropy as a novel signal processing approach: Efficient tool for fault diagnosis of rolling bearing

  • 期刊论文

  • 发表日期：

    2025年3月31日

查看全文
  • Zhe Zhang,
  • Yingwei Liu,
  • Yuxuan Han,
  • Alireza Fatahi-Vajari

摘要

In the practical application of rolling bearing fault information extraction technology, entropy is considered to be a promising feature extraction tool. The traditional entropy methods are vulnerable to environmental noise, which complicates the precise extraction information and results in low fault recognition rates. For this reason, the concept of symbolization is introduced based on entropy theory, using a differential evolution algorithm and an incremental algorithm to enhance method’s performance. the proposed method is called normal cumulative increment arc length entropy (NCIAE). To further obtain more feature information, we suggest extending NCIAE to a multiscale analysis, termed multiscale normal cumulative increment arc length entropy (MNCIAE). Finally, different performance testing models and two operational fault bearing datasets have been used to validate the proposed method. Experimental results demonstrate that this method not only higher computational efficiency but also outperforms traditional methods such as good consistency and robustness, achieving the highest classification accuracy.